Download Polref science case

ISIS Second Target Station Project
ISISTGT2/SAC/02/P4_1
Beamline Name:polREF
Beamline Name: polREF
External Co-ordinator:
Dr. J.P. Goff
Prof. J.A.C. Bland
Dr. S. Langridge
ISIS Contact:
Dr R.M. Dalgliesh
Department of Physics
University of Liverpool
Liverpool L69 7ZE
Tel: 0151 794 3418
Fax: 0151 794 3441
Email:[email protected]
Cavendish Laboratory
University of Cambridge
Cambridge CB3 0HE
01223 337436
01223 350266
[email protected]
The ISIS Facility
Chilton
OX11 0QX
01235 445898
01235 445720
[email protected]
The ISIS Facility
Chilton
OX11 0QX
01235 445687
01235 445720
[email protected]
Summary:
polREF is a polarised neutron reflectometer for studying the inter and intra layer magnetic ordering of surfaces and thin
film systems. These magnetic systems lie at the heart of many of the technological devices that are now being developed in
areas such as computer storage media, field sensing and looking further forward such fields as quantum computation.
Polarised Neutron Reflectometry (PNR) is a technique well matched to the investigation of such systems, providing unique
information in both absolute depth and in-plane magnetometry, as well as imaging. The predicted gain in performance of
polREF (typically a factor ×15-23) over the existing CRISP reflectometer on the first target station, coupled with the new
capability for high-angle scattering, flexible sample environment and advanced beam polarisation conditioning, will
provide a world-class facility, which will lead to new insights in the technologically significant area of thin film
magnetism.
1 Science Case:
1.1 Introduction
It is now possible to grow magnetic multilayers with almost atomic-plane precision, potentially with tailor-made physical
properties. The transition from scientific discovery to commercial exploitation has been dramatic in this field, and the
panoply of practical applications includes magnetic devices for data storage, many types of sensor and, potentially,
quantum computing. The revolutionary development is the creation of devices that exploit the spin rather than the charge
of the electron. One such device made of magnetic multilayers, called a spin valve, is already employed as a read head for
computer hard disks, and this has led to dramatic improvements in data storage densities. Arrays of spin valves may be
used to create magnetic random access memory, which has the advantage of being non-volatile, yet faster, higher density
and lower cost than the current dynamic random access memory devices. The ever-decreasing grain sizes mean that in the
highest density recording media of today are approaching the limits of thermal stability and, as a consequence, patterned
nanoparticle assemblies on surfaces are being considered for data storage. Multilayers also offer a new approach to some
of the canonical problems in condensed matter research, such as the interplay between superconductivity and magnetism,
and quantum confinement within thin layers.
Polarised Neutron Reflectivity (PNR) gives detailed information on the magnetic ordering through a surface or a
buried interface and, as a consequence, it has played a pivotal role in the study of thin-film magnetism. For example,
specular reflectivity determines the magnetic moment magnitude and direction as a function of depth, and it gives
information on magnetic coupling in spin valves and flux penetration in superconductors. Off-specular reflectivity probes
in-plane correlations providing unique insight into the magnetic disorder at interfaces and magnetic interactions in
magnetic nanostructures on surfaces. The polREF reflectometer will offer
dramatic improvements over existing instrumentation and will make a major
contribution to all these areas of research.
FM
1.2 Spinelectronics
1.2.1 GMR Spin Valves
Magnetically coupled multilayers consisting of 3d ferromagnetic (FM) layers
interleaved with non-magnetic spacers (see Fig.1) exhibit giant magnetoresistance (GMR) [1] for appropriate thicknesses of the non-magnetic spacer
layer. These are the regimes in which the oscillatory interlayer coupling has an
antiferromagnetic (AF) ground state between the FM layers. The observed
change in resistivity results from the spin dependent scattering of the conduction
electrons, which depends on the magnetic moment alignment. It is clear that the
GMR will be adversely affected if the formation of a perfect AF coupled
structure is prevented by a vertically incoherent domain structure [2] or a
Page 1 of 6
PNR
NM
FM
ND
AF
Figure 1 A Schematic of a spin valve structure. The
combination of PNR and high angle diffraction will
provide a complete description of the system
ISIS Second Target Station Project
ISISTGT2/SAC/02/P4_1
Beamline Name:polREF
noncolinear arrangement of FM blocks. The ability of the PNR
technique, with polarisation analysis of the reflected beam, to make an
absolute determination of the layer-by-layer magnetic structure allows
very complicated structures to be studied. To optimise devices
knowledge of the magnetic ordering and reversal mechanisms is of key
importance. Real devices tend to be smaller than typical PNR samples,
and the focussing capabilities of polREF will be of great benefit for
these studies.
Recently, the role of the structural interface morphology has
attracted considerable attention because of its important role in the
transport properties [3]. For example, structural roughness on the length
scale of the interlayer magnetic coupling reduces the GMR [4]. Given
the vectorial nature of the magnetisation and the long-range exchange
interaction it is clear that the magnetic disorder need not follow exactly
the structural disorder. A detailed understanding of the interplay
between structural and magnetic morphology is then clearly of real
significance. The magnetic disorder may take the form of magnetic
roughness: the correlation of the structural and magnetic interfaces or
more conventional magnetic domains. Recently, PNR has made
substantial contributions in relating the magnetic and interfacial disorder
Figure 2 The diffuse scattering from a nominal
[Co(9Å)/Cu(9Å)]25. The intensity centred at QZ~0.075Å-1
to the bulk magnetisation and transport measurements [5,6]. Combining
corresponds to the AF ordering wave-vector and arises
the time of flight reflectometry technique with a linear detector
purely from the magnetic scattering. The diffuse
efficiently collects not only the specular reflectivity but also a large
scattering around the AF peak arises from domains.
region of reciprocal space (see Fig. 2) The polREF reflectometer has a
much greater coverage of in-plane wave-vector transfer than CRISP and it will allow better surveys of reciprocal space to
follow short-range correlations. The diffuse magnetic scattering is always weak, and the dramatic increase in signal will be
a huge advantage.
1.2.2 TMR and CMR Devices
In principle, devices with much larger magneto-resistances and, therefore, greater sensitivity may be produced. For
example, a tunnel junction may be obtained by replacing the non-magnetic spacer layer (as would be found in a GMR
device) with an insulating layer. The tunnelling current through the insulating layer is very sensitive to the mutual
orientation of adjacent FM layers, and large tunnelling magneto-resistances (TMR) have been reported. Unfortunately
such results are strongly temperature dependent, possibly due to misaligned spins at the interface. If such TMR devices are
to be commercially realised such difficulties need to be overcome. Colossal magneto-resistance (CMR) has generated
enormous experimental interest due to the extremely large change in magneto-resistance associated with the metalinsulator transition in systems such as the manganites. Moreover, the conduction bands of such materials are fully spin
polarised leading to exciting prospects for spin injection devices. At the moment CMR position and timing sensors are
used to reduce emissions from car engines. The performance of such devices, as with tunnelling, is critically dependent on
the magnetic properties of the interfacial region. We expect that polREF will make significant contributions to these new
areas for the reasons outlined above for GMR devices.
1.2.3 Exchange Bias
Exchange bias, in which hysteresis loops are shifted from the zero-field axis, has been observed in a variety of systems
comprising FM materials in contact with AF materials. The origin of this phenomenon is the interfacial exchange
interaction between the FM and AF layers, but there are many unresolved issues. For example, the biasing fields are much
smaller than expected, and it is a long-standing puzzle why exchange bias is observed at all for interfaces that are rough on
an atomic scale. These issues are of enormous technological importance since GMR spin valves and tunnel junctions rely
on the exchange biasing of one of the FM layers to magnetically pin this layer. High-angle neutron diffraction (ND) on
polREF will allow the investigation of the AF ordering in epitaxial systems (See Fig. 1).
1.2.4 Spin-injection in Semiconductors
Although many hybrid FM/semiconductor device designs have been proposed, finding materials with which to implement
them is a significant experimental challenge. Desirable properties include half-metallicity, and PNR allows the degree of
spin polarisation to be directly determined in the interfacial region. Silicon-based spin transistors are being developed
using different tunnel barriers in order to deal with the conductivity mismatch at the metal-semiconductor interface. The
increased flux and dynamic range of polREF will enable the interfacial region to be studied with a new degree of statistical
accuracy. For example, a non-uniform magnetisation profile, such as reduced magnetisation in the near interface region,
would equate to a reduced polarisation and therefore a reduction in the spin injection.
1.3 Magnetic Nanostructures
1.3.1 Magnetic Dot Arrays
In the search for ever higher recording densities conventional recording media are approaching the so-called
superparamagnetic limit as the size of the data bits is becoming increasingly small. As we approach this limit, the stored
Page 2 of 6
ISIS Second Target Station Project
ISISTGT2/SAC/02/P4_1
Beamline Name:polREF
Magnetization (arbitrary scale)
data becomes thermally unstable and it is unreliable. To increase storage densities to the order of Tera-bits/square inch
(1012bits/in2) requires new approaches using nanotechnology. One promising idea is to pattern laterally magnetic films.
Several techniques are currently under consideration for the self-assembly of regular arrays of 0D magnetic islands. Such
patterned elements can only support two magnetic states, and the resulting small size features lend themselves to extremely
high recording densities. To fully understand and optimise such systems requires a full knowledge of the magnetisation
density within the magnetic islands and the nature of any magnetic interactions between the dots. polREF will provide
such information on both the structural and magnetic order through the harmonics of the relectivity. It is increasingly
recognised that the interplay between the physical and magnetic structure can significantly influence the device
performance. Studying how the patterned magnetic structure responds to external influences (see sample environment)
such as temperature and magnetic field and intrinsic parameters such as the size and shape of the elements will lead to
greater insight into these challenging nanomaterials. Given the complexity of such experiments the literature contains few
examples of such studies [7,8]: a scarcity that the capabilities of polREF will address. Such measurements will
complement those eventually possible on the offSPEC beam line utilising the more difficult spin echo technique, which
potentially gives full two-dimensional information on the in-plane correlations.
Although the in-plane correlations are projected onto one dimension for polREF
essential information on the relation between structure and magnetism and magnetic
interactions will be available immediately.
1.3.2 Magnetic Stripes
Another aspect of nanotechnology where polREF will be able to make an immediate
impact is the study of magnetic nanowires (see Fig. 3). One-dimensional magnetic
stripes may readily be grown at the step edges on miscut substrates. In this case the
geometry of polREF is ideally suited to measure the Bragg peaks from the wires,
enabling a detailed comparison between structure and magnetism and a study of the
development of magnetic interactions. [In principle, the magnetic stripes should not
Figure 3 Fe stripes at the step
exhibit long-range magnetic order at all, but the magnetic interactions between the
edge of miscut Si.
stripes are believed to be the key ingredient leading to ordering.]
1.4 Fundamental Magnetism
1.4.1 Superconductivity
There has recently been much interest in the interplay between magnetism and superconductivity. For example, knowledge
of the magnetic induction profile at the surface of a superconductor allows one to test ideas in the Ginzburg-Landau
theory. PNR has been demonstrated to provide valuable information
on such parameters as the penetration depth [9], which are less readily
Normal
accessible through other techniques. Furthermore, PNR has been used
to study magnetic vortices, which form in type-II superconductors
[10]. The current transport in such systems, which is crucial to
0.008
technological applications, depends on the pinning of such vortices.
FC
0.006
Moreover, it is well known that the properties of thin film
superconductors can be modified when they are in contact with a
0.004
ferromagnetic material: the proximity effect. PNR is able to provide
0.002
information on both the flux penetration and the magnetic structure.
ZFC
The coexistence of 3D superconductivity and magnetism has recently
0.000 S
been discovered at ISIS [11]. The superconducting ordering
-0.002
temperature of blocks much thinner than the coherence length is
comparable to the bulk for an antiferromagnetic alignment of
0
100
200
300
ferromagnetic blocks, but the superconductivity is suppressed when
Temperature
(K)
the orientation is ferromagnetic, suggesting new possibilities to study
Superconducting
the interplay between these usually mutually exclusive phenomena
and the potential for a superconducting spin valve (See Fig. 4). The
enhanced flux, coupled with a greater range of external parameters
Figure 4 polREF will allow the study of proximity effects in
(temperature, magnetic field) on polREF will allow data collection of
superconducting/magnetic systems.
high statistical quality (essential for the study of such subtle
phenomena) and will provide greater insight into the ordering
mechanisms in such systems.
1.4.2 Model Magnetic Systems
The influence of the surface on phase transitions is a problem of fundamental importance in statistical physics. Despite
considerable progress in theoretical studies of surface magnetic phase transitions, attempts to verify the predictions
experimentally have been thwarted by the severe demands on surface quality. However, in magnetic multilayers one can
study the first few blocks of magnetic material rather than the first few atomic planes. The surface spin-flop transition
predicted for a chain of XY spins with AF interactions and uniaxial anisotropy was recently detected using PNR [12]. In a
similar approach a ferromagnetically coupled multilayer grown on an antiferromagnetically coupled multilayer has been
Page 3 of 6
ISIS Second Target Station Project
ISISTGT2/SAC/02/P4_1
Beamline Name:polREF
used to shed light on the mechanism of exchange bias [13] (Fig. 5). Many other problems can be approached in this
manner, and polREF will provide the sensitivity required to study weak features such as domain walls.
INTENSITY
1.4.3 Lanthanide and Actinide Research
The large magnetic moments and anisotropies in lanthanide systems make them of great interest. An AF neutron
diffraction peak has been observed from a Ho film only 46Å thick using ADAM at the ILL [14]. The high-angle
diffraction capability of polREF will give wave vector information on wavefunctions in thin films, providing new
opportunities to study quantum size effects. Given the large anisotropy within these systems the provision of a high
magnetic field (>10T) cryomagnet at low temperatures (<2K) will vastly extend the parameter space available for such
studies. The actinide elements exhibit behaviour somewhere in between those of the transition metals and the rare earths.
Surprisingly, uranium based multilayers have been considered as
10-1
candidate technologies for high density recording media due to the
enormous magneto-optical effect. The quantities of radioactive
materials in such samples are very small (typically <100µg) thereby
10-2
minimising the radioprotection requirements. Such studies are still
difficult and limited due to sample preparation but these issues are
10-3
being actively addressed and such topics as induced magnetism,
hydrogen loading etc. in actinide systems will offer new research
10-4
possibilities.
1.5 Additional areas of research and future potential
10-5
1.5.1 Neutron Depolarisation
The ability to study the three dimensional structure of
ferromagnetic domains is of relevance in high performance metallic
10-6
0.04
0.06
0.08
0.10
0.12
0.14
alloys and in the optimisation of neutron polarising mirrors. The
technique [15] involves measuring the 9 (XYZ) spin dependent
Q(Å-1)
cross-sections. The flexible beam polarisation on polREF will
Figure 5 Direct observation of a spin flop phase from PNR data in
a model exchange bias system.
facilitate such measurements, which when coupled with the large
gain in performance will allow more parametric and angular
dependence studies.
1.5.2 The Inverse Phase Problem
In principle, a complete determination of the complex reflection coefficients in a reflectometry experiment is possible
using magnetic reference layers [16]. This is commonly referred to as the inverse phase problem and is well known in both
neutron and x-ray scattering communities. The addition of an additional scattering potential (the magnetic reference layer)
allows one to solve uniquely the scattering potential profile, which gave rise to the observed reflectivity. Experimentally,
such ideas are complicated through a finite wave-vector range, statistics and the need for a complete characterisation of the
reference layer. Clearly, polREF will significantly contribute in wave-vector range and counting statistics. Such techniques
offer promise for complicated biological membrane structures where a-priori knowledge is not readily available [17].
1.5.3 In-situ growth
An exciting possibility for polREF is the addition of a UHV deposition chamber at the sample position. Many of the recent
advances in magnetic nanostructures have been driven by developments in deposition techniques such as sputtering and
molecular beam epitaxy (MBE). The ability to study magnetic surfaces during in-situ growth with reflectometry would
provide a new and unique insight into the growth and magnetic ordering mechanisms in magnetic thin film systems. The
avoidance of surface contamination or capping layers can be crucial in, for example, understanding finite size effects [18].
2 Technical Case:
The design of polREF is such that it dramatically extends the measurements currently possible on existing polarised
neutron reflectometers. It clearly complements the science case presented for the offSPEC proposal in the study of in-plane
order/disorder and patterned substrates, and the design also allows surface chemistry experiments of the type discussed in
the INTER case.
2.1 Outline Technical Specification:
The outline technical design of polREF has been developed to optimise the possibilities for PNR on the ISIS second target
station in the following key areas:
• Enhanced Flux at the sample position
• Maximise use of the increased wavelength band
• Minimise the noise floor on the instrument through design and shielding
• Secondary shutter to reduce access time
• Flexible neutron polarisation control
Page 4 of 6
ISIS Second Target Station Project
ISISTGT2/SAC/02/P4_1
Beamline Name:polREF
•
Optimised sample environment
These points are addressed in the following sections. The performance summary is presented in table 1.
Moderator
Coupled grooved cold H2/CH4
Incident Wavelength
0.9Å - 15Å
Resolution
~3-10%∆Q/Q
Flux at the sample position
~107 n cm-2 s-1
L1
23m
L2
3m
Beam size
Performance
60 mm x 30mm H x V
×15-23 over ISIS Target Station
Table 1 Outline design specification for the polREF reflectometer
2.2 Neutron collection and transport: optics
The beamline optics (see Fig. 6) will
consist of a channelled beam bender and a
tapered guide. The bender will bend the
neutron beam down by 2.3° (ensuring
compatibility with the inter reflectometer
proposal for liquid surfaces, ferrofluids
etc.) and minimising the fast neutron and γray background from the direct line of sight
with the moderator. The bender design is
being optimised but is likely to consist of a
long bender with multiple channels. An
interesting possibility is that the bender
Figure 6 An Outline layout for polREF
may be mounted in the primary shutter although this
represents an engineering challenge to ensure repeatable
positioning. A secondary fast shutter will be used to control
the neutron access to the blockhouse. Long wavelength
neutrons will be removed through a frame overlap mirror.
10
1
0.02
0.01
0.001
0.015
0.0001
0.01
0.00001
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
Qz(Å-1)
Reflectivity
0.1
0.005
-1
Qz (Å )
0
Figure 7 Comparison of the accessible wave-vector range on CRISP in two
angular settings (bottom) against polREF (top) . The curves have been offset
for clarity.
-0.005
-0.01
Samples for PNR experiments are typically much
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
-1
Qx(Å )
smaller (~1cm2) than those encountered in surface chemistry
(~40cm2) for example. In addition, there is poor resolution in
Figure 8 The accessible wave-vector range on CRISP (red) for a typical
the horizontal plane (cf. offspec). One can then use this to
experimental geometry and polREF (black).
gain in neutron flux by tapering the guide in the horizontal
direction at the expense of the wavelength dependent
divergence. Monte Carlo simulations are currently underway to calculate the acceptable horizontal divergence without
compromising the vertical resolution. To minimise scattering and absorption the system will be kept under vacuum and
windows kept to a minimum. At the current state of design, the relative merits of background suppressing and wavelength
defining choppers against the loss in short wavelength flux are being assessed. The large gains in wave-vector range both
Page 5 of 6
ISIS Second Target Station Project
ISISTGT2/SAC/02/P4_1
Beamline Name:polREF
specular and diffuse coupled with the optimised flux will lead to a large increase in data collection efficiency. This will
allow one to study smaller samples (an increasing requirement for patterned samples), more complex sample environment
or to survey reciprocal space.
2.3 Polarising and Flipping Elements:
In the wavelength range (1-14Å) of interest on polREF, polarising supermirrors match the requirement for a high
polarisation of >95%, high reflectivity >80% and transmission >90%. 80% reflectivity up to m=4 is now realisable. To
simplify the subsequent beamline the polariser will be run in a transmission mode with the flexibility to run in reflection
should there be new polariser designs, background minimisation issues to be addressed.
For the analysing assembly it is clear that the ability to analyse a diffuse beam is essential. A curved stacked
supermirror assembly (oriented perpendicular to the incident polariser) provides a high degree of polarisation with good
angular coverage. The design is finalised but may take advantage of the solid state polarisers developed at HMI [19].
The flipping elements are characterised by a high flipping efficiency (>99.9%) over the entire wavelength range
and over the beam size (40mm). This can be realised by several technologies relying on either adiabatic or non-adiabatic
rotation of the neutron spin. In addition, the flipper should not introduce significant background and be tolerant to stray
magnetic field (associated with the cryomagnet for example). Technologies currently under investigation include Drabkin
spinflippers and adiabatic rotators [20] both of which are in use on the CRISP reflectometer at ISIS. The flippers will also
be used to control the incident polarisation onto the sample allowing us to access all the elements of the polarisation
matrix. Permanent magnets will provide the neutron guide field. Design effort will be given to minimising the effects of
stray magnetic fields on the polarising and guide field elements and the need for active compensation control.
2.4 Sample Environment:
The science case determines the required sample environment, which consists of primarily two regimes. For many
transition metal systems room temperature and low magnetic fields are required. Of more importance is the accurate
control of the applied magnetic field. This will be achieved through a computer controlled bipolar electromagnet
(µoH<1.5T). In parallel, in-situ Magneto-Optic Kerr microscopy will provide detailed information on the magnetic
configuration complementing and maximising the efficiency of the neutron measurements. The rare earth and
superconductivity measurements are characterised by low temperatures (<2K) and high magnetic fields (~10T). These can
be realised using a superconducting magnet. In addition, the proposal requires a magnetic field (~0.5T), which can be
applied in any direction preferably with a maximum vertical field to pull in-plane magnetism out of the plane. Such a
magnet is matched to the controllable incident polarisation allowing access to the full elements of the polarisation matrix.
All of the beamline sample environment will be mountable on heavy duty (100Kg load capability) high precision stages.
2.5 Neutron Detection:
For high count rate, high signal-to-noise ratio measurements a 3He single detector will be required. For the off-specular
and out of plane diffraction measurements a 200mm×200mm area detector with a resolution of <1mm. With a detector
distance of 3m this represents an angular resolution of ~0.02° which is well matched to the upstream resolution. To study
the structural and magnetic ordering on an atomic length scale polREF will require a high angle detector capability. To
maximise the intensity from the broad reflections, typical of thin film samples relaxed collimation will be employed.
[1] M.N. Baibich et al. Phys. Rev. Lett. 61, 2472 (1988)
[2] H. Holloway and D.J. Kubinski, J. Appl. Phys. 83, 2705 (1998)
[3] S.S.Parkin, Ultrathin Magnetic Structures II, Springer-Verlag, Berlin (1994)
[4] Z.P. Shi et al. Phys. Rev. B. 49, 15159 (1994)
[5] C. Leighton et al. Phys. Rev. Lett. 86 4394 (2001)
[6] S. Langridge et al. Phys. Rev. Lett. 85 4964 (2000)
[7] K. Temst et al. J.M.M.M. 226-230 1840 (2001)
[8] N.D. Telling et al. J. Appl. Phys. (in press), ISIS Annual Report highlight 2002
[9] M.P. Nutley et al. Phys. Rev. B. 49 15789 (1994)
[10] S-W. Han et al. Phys. Rev. B. 59 14692 (1999)
[11] J.P. Goff et al. J.M.M.M. 240 592 (2002)
[12] S.G.E. te Velthius et al. Phys. Rev. Lett. 89 127203 (2002)
[13] P. Steadman et al. Phys. Rev. Lett. 89 077201 (2002)
[14] V. Leiner et al. Physica B 283 167 (2000)
[15] W.H. Kraan et al. J. Mag. Mag. Matl. 236 302 (2001)
[16] Leeb et al. Phys. Rev. B. 63 045414 (2001)
[17] Majkrzak et al. Biophysocal Journal 79 3330 (200)
[18] H. Yamazaki et al. J.M.M.M. 198-199 279 (1999)
[19] Th. Krist, F. Mezei. Physica B 276 208 (2000)
[20] See for example T.J.L. Jones et al. Nucl. Instr. And Meth. 152 463 (1978)
Page 6 of 6